Introduction

Triple negative breast cancer has very poor prognosis (Leon-Ferre, 2022). While immunetherapy with immune checkpoint inhibitors has improved survival outcomes, the benefit is limited to a small subset of patients (10-15%) (Abdou Y., et al, 2022). Thus, there is a need to develop effective therapeutic approaches for the treatment of TNBC. Tumor microenvironment of TNBC consists of immune suppressive myeloid populations such as TAMs and MDSCs which hinder anti-tumor CD8+ T cell responses . Moreover, tumor induces the generation and recruitment of TAMs and MDSCs. Identification of mechanisms that mediate crosstalk between tumor and immune cells will provide actionable therapeutic targets. Our group has identified that the activity of p38 alpha (p38a) MAP kinase in the tumor cells promotes the generation and recruitment of these populations (Rajan P., et al, 2025, under review at Can Imm Res). In this study, we evaluated the effect of p38a blockade using pharmacological inhibition and genetic depletion on the tumor immune landscape in the 4T1 mouse mammary carcinoma model. Live CD45+ immune populations were flow sorted from 4T1 tumors treated with vehicle (4T1-Tu), p38 inhibitor LY2228820 (4T1-Tu-plus-p38i) or 4T1 tumors with p38a knockout (p38a-ko-H9) and single cell RNA-seq was performed using the 10X Chromium 3’ RACE technology. CellRanger was used to align reads to mouse genome build mm10 and generate read-count matrices. In this section (Part 1), quality control is performed to filter out low quality cells and cells are visualized in 2D using dimensionality reduction algorithms.

Setup:

knitr::opts_chunk$set(echo = TRUE, cache = FALSE)

#libraries
library(scater)
library(scran)
library(SingleCellExperiment)
library(ggplot2)
library(ComplexHeatmap)
library(circlize)
library(Matrix)
library(knitr)
library(kableExtra)
library(RSpectra)
library(dplyr)
library(celldex)
library(SingleR)

# colors
palette(c("dodgerblue","tomato","forestgreen","orange","orchid",
          "darkslategray2","firebrick","lightgreen","gold","hotpink3",
          "lightblue","deeppink","purple","darkorange","plum1"))

Reading input data:

#Load data
path <- "./Bakin/GSRdata/"
files <- list.files(path)
files <- files[c(1,2,3)]
samples <- gsub("^.{12}|.{18}$","",files)

sce <- vector("list", length(samples))
names(sce) <- samples

for (i in 1:length(samples)){
  matrix_dir = paste0(path, files[i],"/outs/filtered_feature_bc_matrix/")
  cellbarcodes <- read.table(paste0(matrix_dir, "barcodes.tsv.gz"))
  genenames <- read.table(paste0(matrix_dir, "features.tsv.gz"))
  countsmat <- Matrix::readMM(paste0(matrix_dir, "matrix.mtx.gz"))
  rownames(countsmat) <- genenames[,2]
  colnames(countsmat) <- cellbarcodes[,1]
  samplename <- data.frame(samplename = rep(samples[i], ncol(countsmat)))
  sce[[i]] <- SingleCellExperiment(assays = list(counts = as.matrix(countsmat)),
                                 colData = samplename)
  colnames(sce[[i]]) <- paste(samples[i], cellbarcodes[,1], sep="_")
}
rm(list = ls()[!grepl("my|samples|sce", ls())])
#png function
pngfunc <- function(filename = "mypng.png", width=7, height = 7, units = "in",
                    res = 300, ...){
  png(filename = filename, width = width, height = height, units = units,
      res = res, ...)
}

Per Cell QC

Cells that meet the following criteria will be retained for further analysis * > 1500 read counts * > 300 detected genes * < 20% mitochondrial content

#Using the addPerCellQC function from scater package to compute quality contorl metrics.
sce <- lapply(sce, function(x) {
  x <- addPerCellQC(x, subsets = list(Mito = grep("^mt-", rownames(x)),
                                      Ribo = grep("^Rp[l|s]", rownames(x)),
                                      Hemo = grep("^Hb[a|b]", rownames(x))))
  x
})

#Determining which cells pass the criteria
sce <- lapply(sce, function(x) {
  x$qc_count <- x$sum > 1500
  x$qc_gene <- x$detected > 300
  x$qc_mito <- x$subsets_Mito_percent < 20
  
  x$qc_pass <- x$qc_count & x$qc_gene & x$qc_mito
  x
  })

#Generating QC plots
pngfunc("Plots/QC2/percellQC.png", height = 7, width = 9)
layout(t(matrix(c(0:19),4,5)), widths = c(3,3,3,3), heights = c(1,3,3,3,1))
par(mar = c(0,0,0,0))
plot.new(); text(0.5, 0.5, "No. of reads", cex = 2)
plot.new(); text(0.5, 0.5, "No. of genes", cex = 2)
plot.new(); text(0.5, 0.5, "% Mitochondrial genes", cex=2)
for(i in samples){
  par(mar = c(0,0,0,0)) 
  plot.new(); text(0.5, 0.5, i , cex=1.5)
  par(mar = c(2,2,0,0.5))
  
  hist(log10(sce[[i]]$sum), breaks=200, main="",
       xlim=c(2.5,5.5), xlab="", ylab="")
  abline(v=log10(1500), col=2)
  
  hist(log10(sce[[i]]$detected), breaks=200, main="",
       xlim=c(1,4), xlab="", ylab="")
  abline(v=log10(300), col=2)
  
  hist(log10(sce[[i]]$subsets_Mito_percent), breaks=200, main="",
       xlim=c(0,100), xlab="", ylab="")
  abline(v=20, col=2)
  }
par(mar = c(0,0,0,0))
plot.new()
plot.new(); text(0.5,0.5,"log10 total reads")
plot.new(); text(0.5,0.5,"log10 total genes")
plot.new(); text(0.5,0.5,"% Mitochondria")

invisible(dev.off())

include_graphics("Plots/QC2/percellQC.png", dpi=300)

#Generating QC Table
QC <- c()
for(i in samples){
  QC <- rbind(QC, as.data.frame(colData(sce[[i]])))
}

df <- data.frame(Sample = samples)
df[,c("bqc","genes", "reads", "mito", "total", "aqc", "ribo")] <- NA

for(i in df$Sample){
  x<- QC[QC$samplename==i, ]
  df$BQC[df$Sample==i] <- nrow(x)
  df$genes[df$Sample==i] <- as.integer(sum(!x$qc_gene) / nrow(x) * 100)
  df$reads[df$Sample==i] <- as.integer(sum(!x$qc_count) / nrow(x) *100)
  df$mito[df$Sample==i] <- as.integer(sum(!x$qc_mito) / nrow(x) * 100)
  
  df$total[df$Sample==i] <- as.integer(sum(!(x$qc_gene & x$qc_count & x$qc_mito)) /
    nrow(x) * 100)
  df$AQC[df$Sample==i] <- sum(x$qc_pass)
}

df[,1:7] %>%
  kable(format = "html", align = "c", escape = F) %>%
  kable_styling("striped")
Sample bqc genes reads mito total aqc
4T1-Tu NA 1 15 3 16 NA
4T1-Tu-plus-p38i NA 0 24 1 24 NA
p38a-ko-H9 NA 4 23 8 25 NA 
#Combine all data into one sce object
assign("sce", do.call("cbind", sce))

Per Gene QC

The counts are normalized to library size and converted to counts per million units in log2 scale. Ideally we would like to analyze genes that vary upon p38 inhibition by p38i or p38KO as those are the most interesting changes. Hence, genes that vary little between samples are removed.

#Normalizing counts and converting to log2 counts per million
cpm_mat <- as.matrix(counts(sce))
lib_size <- colSums(cpm_mat)
cpm <- t(t(cpm_mat/lib_size)*1000000)
cpm <- cpm + 1
log2cpm <- log2(cpm)
  
assay(sce, "cpm") <- cpm
logcounts(sce) <- log2cpm
  
#Calculating which genes show variance greater than 0.5
keep <- rowVars(logcounts(sce)) > 0.5

#Generating per Gene QC plots
pngfunc("Plots/QC2/pergeneQC.png", width = 5, height=4)
layout(t(matrix(c(0:5),2,3)), widths = c(1,4), heights = c(1,4,2))
par(mar=c(0,0,0,0)) 
plot.new();text(0.5,0.5, "Mean vs Variance plot", cex=1.5)
plot.new(); text(0.5,0.5,"Variance of Gene counts (log2CPM)", srt = 90, cex=1.2)
par(mar = c(2,2,3,1))
plot(rowMeans(logcounts(sce)), rowVars(logcounts(sce)), xlab = "",
       ylab = "", 
     col = ifelse(rowVars(logcounts(sce))>0.5, "orange", "blue"), pch=19, cex=0.5)

plot.new();
plot.new(); text(0.5, 0.5, "Mean Gene counts (log2CPM)", cex = 1.2)

invisible(dev.off())

include_graphics("Plots/QC2/pergeneQC.png", dpi=300)

#Generating per gene QC table
df_gene <- data.frame(Sample=samples)
df_gene[,c("BQC","P_Filtered", "AQC")] <- NA
df_gene$BQC <- nrow(sce)
df_gene$P_Filtered <- as.integer(sum(!keep) / nrow(sce) * 100)
df_gene$AQC <- sum(keep)

df_gene[,1:4] %>%
  kable(format = "html", align = "c", escape = F) %>%
  kable_styling("striped")
Sample BQC P_Filtered AQC
4T1-Tu 32285 65 11142
4T1-Tu-plus-p38i 32285 65 11142
p38a-ko-H9 32285 65 11142 
#Filtering sce object based on QC criteria
sce <- sce[keep, ]

Visualizing in 2D space

Principal component analysis (PCA) is a commonly used method to compact data and assess distribution of data points. The runPCA() function determines PCs using the first 500 genes that show the highest variation. t- distributed stochastic neighborhood embedding is used for more complex single cell datasets. Here, I ran t-SNE using the pre-existing PCA results as input to the tSNE algorithm. This is useful as it improves speed by using a low-rank approximation of the expression matrix; and reduces random noise, by focusing on the major factors of variation.

#Determining tSNE dimensions
set.seed(100)
sce <- runPCA(sce)
sce <- runTSNE(sce, dimred = "PCA", perplexity = 40)

sce$TSNEx <- reducedDims(sce)$TSNE[,1]
sce$TSNEy <- reducedDims(sce)$TSNE[,2]

pngfunc(filename = "Plots/QC2/tSNE.png", width= 9, height = 7)
layout(matrix(c(1,4,2,5,3,6),2,3), widths = c(3,3,3))
par(mar = c(1,1,2,1))
plot.new()
text(0.5,0.5, "All samples combined", cex=2)
plot(sce$TSNEx, sce$TSNEy, pch = 19, cex = 0.2, xlim=c(-40,40),col = factor(sce$samplename), 
     xlab = "", ylab = "", xaxt = "n", yaxt = "n", main = "Cell types in all samples combined")
plot.new()
legend("center", legend = unique(sce$samplename), col = palette(), bty="n", ncol = 1, pch = 19, 
       title = "Samples", cex = 2)
for(i in unique(sce$samplename)){
  a <- ifelse(sce$samplename == i, 1, 0)
  b <- order(a)
  c <- ifelse(sce$samplename == i, which(unique(sce$samplename)==i), "gray")
  par(mar = c(1,1,2,1))
  plot(sce$TSNEx[b], sce$TSNEy[b], pch = 19, cex = 0.2, xlim=c(-40,40),
       col = c[b], 
       xlab = "", ylab = "", xaxt = "n", yaxt = "n", main = i)
}

invisible(dev.off())

include_graphics("Plots/QC2/tSNE.png", dpi = 300)

Annotation using ImmGen data

The immune populations present in the samples were identified using SingleR. SingleR uses a reference dataset containing known labels. It then labels each cell based on similarity to the reference. This is done by comparing the expression of a set of marker genes and using spearman correlation to assign a label to each cell. Immunological genome database containing microarray based gene expression data for multiple mouse immune cell populations was used as a reference. This was accessed through the celldex package.

#cell type annotation using reference dataset from ImmGen db

ref1 <- celldex::ImmGenData()
pred1 <- SingleR(test=sce, ref=ref1, labels=ref1$label.main)

#assigning cells with their cell types
sce$immgenlabel <- pred1$labels

#heatmap to which label received the highest assignment score in each cell
pngfunc("Plots/QC2/scoreheatmap.png", width = 11, height = 7)
plotScoreHeatmap(pred1)
invisible(dev.off())
include_graphics("Plots/QC2/scoreheatmap.png", dpi = 300)

#TSNE plots

#cell types identified overall
pngfunc("Plots/QC2/cell_types.png", width = 7, height =7)
layout(matrix(c(1,2,0,3),2,2), widths = c(4,2), heights = c(2,4))

plot.new(); text(0.5, 0.5, "Cell types identified overall", cex = 2)
par(mar=c(1,1,1,0))
plot(sce$TSNEx, sce$TSNEy, xlab = "", ylab = "", xaxt = "n", yaxt = "n",
     col = factor(sce$immgenlabel), pch = 19, cex = 0.2)
plot.new(); legend("center", legend = levels(factor(sce$immgenlabel)), ncol = 2, cex = 1, title = "Cell types",
       pch = 19, col = palette(), bty = "n")
invisible(dev.off())
include_graphics("Plots/QC2/cell_types.png", dpi = 300)

#cell types identified in each sample
pngfunc("Plots/QC2/cell_types_each.png", width = 6, height = 9)
layout(matrix(c(1,3,5,2,4,6), 3,2), widths = c(3,2), heights = c(3,3,3))
par(mar = c(0.5,1,2,1))
for(i in unique(sce$samplename)) {
  s <- sce[ ,sce$samplename==i]
  plot(s$TSNEx, s$TSNEy, xlab="",ylab="",xaxt="n",yaxt="n",
       main = i, col = factor(s$immgenlabel), pch = 19, cex=0.2)
  plot.new(); legend("center", legend = levels(factor(s$immgenlabel)), ncol = 2, cex = 1,
                     title = "Cell Types",
                     pch=19, col= palette())
}
invisible(dev.off())
include_graphics("Plots/QC2/cell_types_each.png", dpi = 300)

#Create tables with cell type percentages
mylist <- vector("list", length(samples))
names(mylist) <- samples
for (i in 1:length(samples)){
  df <- table(sce[ ,sce$samplename==samples[i]]$immgenlabel)
  df <- as.data.frame(df)
  df$Percent <- df$Freq/sum(df$Freq) * 100
  df$Sample <- rep(samples[i], nrow(df))
  names(df)[1] <- "Cell_Type"
  mylist[[i]] <- df
  }

#Combine 3 lists onto one data frame
DF <- Reduce(function(...) merge(..., all=TRUE), mylist)

#Barplot of cell type percentages

pngfunc("Plots/QC2/Cell_types_Percent.png", width = 11, height  = 6)
ggplot(DF, aes(fill=Sample, x=Cell_Type, y=Percent)) +
  geom_bar(position="dodge", stat="identity")+
  geom_hline(yintercept = 2, color = "black", linetype = "dashed") +
  scale_x_discrete(guide = guide_axis(angle = 45)) +
  ggtitle("Proportions of cell types") +
  theme_bw() +
  scale_y_continuous(limits=c(0,100))

invisible(dev.off())
include_graphics("Plots/QC2/Cell_types_Percent.png")

Subset sce to retain cell types present more than 1%

Analysis of cell type percentages revealed that the following cell types are more than 1% abundant. These will be further analysed while the rest will be filtered out

types <- c("DC", "T cells", "Monocytes", "Macrophages", "Neutrophils",
            "NKT", "NK cells", "ILC")
sce <- sce[ ,sce$immgenlabel %in% types]

#Create tSNE plots of subsetted sce object
pngfunc("Plots/QC2/cell_types_after_subsetting.png", width = 6, height = 9)
layout(matrix(c(1,3,5,2,4,6), 3,2), widths = c(3,2), heights = c(3,3,3))

par(mar = c(0.5,1,2,1))
for(i in unique(sce$samplename)) {
  s <- sce[ ,sce$samplename==i]
  plot(s$TSNEx, s$TSNEy, xlab="",ylab="",xaxt="n",yaxt="n",
       main = i, col = factor(s$immgenlabel), pch = 19, cex=0.2)
  plot.new(); legend("center", legend = levels(factor(s$immgenlabel)), ncol = 2, cex = 1,
                     title = "Cell Types",
                     pch=19, col= palette())
}
invisible(dev.off())
include_graphics("Plots/QC2/cell_types_after_subsetting.png", dpi = 300)

#Create tables with cell type percentages
mylist <- vector("list", length(samples))
names(mylist) <- samples
for (i in 1:length(samples)){
  df <- table(sce[ ,sce$samplename==samples[i]]$immgenlabel)
  df <- as.data.frame(df)
  df$Percent <- df$Freq/sum(df$Freq) * 100
  df$Sample <- rep(samples[i], nrow(df))
  names(df)[1] <- "Cell_Type"
  mylist[[i]] <- df
}

#Combine 3 lists onto one data frame
DF <- Reduce(function(...) merge(..., all=TRUE), mylist)

#Barplot of cell type percentages
pngfunc("Plots/QC2/Cell_types_subsetted_Percent.png", width = 11, height  = 6)
ggplot(DF, aes(fill=Sample, x=Cell_Type, y=Percent)) +
  geom_bar(position="dodge", stat="identity")+
  ggtitle("Proportions of cell types") +
  theme_bw() +
  scale_y_continuous(limits=c(0,100))

invisible(dev.off())
include_graphics("Plots/QC2/Cell_types_subsetted_Percent.png")

Plotting cell types based on markers

Feature plots are a convenient way to represent distinct cell types as well as their distribution relative to other cell types in tSNE space. Here, two gene markers (features) representing each of the immune populations is shown in red and blue. Combination of red and blue gives purple, highlight the specific cell population.

#Feature plots
rgb <- vector("list", length = 8)
names(rgb) <- types
for(i in types){
  if(i == "T cells"){
    cd8 <- as.numeric(logcounts(sce["Cd8a",]))
    cd3d <- as.numeric(logcounts(sce["Cd3d",]))
    cd8 <- cd8/max(cd8)
    cd3d <- cd3d/max(cd3d)
    rgb[[i]] <- rgb(cd8,0,cd3d)
  } else if(i == "Monocytes"){
    ccr2 <- as.numeric(logcounts(sce["Ccr2",]))
    ly6c2 <- as.numeric(logcounts(sce["Ly6c2",]))
    ccr2 <- ccr2/max(ccr2)
    ly6c2 <- ly6c2/max(ly6c2)
    rgb[[i]] <- rgb(ccr2,0,ly6c2)
  } else if(i == "Neutrophils"){
    csf3r <- as.numeric(logcounts(sce["Csf3r",]))
    g0s2 <- as.numeric(logcounts(sce["G0s2",]))
    csf3r <- csf3r/max(csf3r)
    g0s2 <- g0s2/max(g0s2)
    rgb[[i]] <- rgb(csf3r,0,g0s2)
  } else if(i == "DC"){
    p2ry10 <- as.numeric(logcounts(sce["P2ry10",]))
    itgax <- as.numeric(logcounts(sce["Itgax",]))
    p2ry10 <- p2ry10/max(p2ry10)
    itgax <- itgax/max(itgax)
    rgb[[i]] <- rgb(p2ry10,0,itgax)
  } else if(i == "Macrophages"){
    arg1 <- as.numeric(logcounts(sce["Arg1",]))
    c1qa <- as.numeric(logcounts(sce["C1qa",]))
    arg1 <- arg1/max(arg1)
    c1qa <- c1qa/max(c1qa)
    rgb[[i]]<- rgb(arg1,0,c1qa)
  } else if(i == "NK cells"){
    nkg7 <- as.numeric(logcounts(sce["Nkg7",]))
    gzmb <- as.numeric(logcounts(sce["Gzmb",]))
    nkg7 <- nkg7/max(nkg7)
    gzmb <- gzmb/max(gzmb)
    rgb[[i]] <- rgb(nkg7,0,gzmb)
  } else if(i== "NKT"){
    icos <- as.numeric(logcounts(sce["Icos",]))
    lck <- as.numeric(logcounts(sce["Lck",]))
    icos <- icos/max(icos)
    lck <- lck/max(lck)
    rgb[[i]] <- rgb(icos,0,lck)
  } else if(i == "ILC"){
    tbx21 <- as.numeric(logcounts(sce["Tbx21",]))
    ncr1 <- as.numeric(logcounts(sce["Ncr1",]))
    tbx21 <- tbx21/max(tbx21)
    ncr1 <- ncr1/max(ncr1)
    rgb[[i]]<- rgb(tbx21,0,ncr1)
  } else {
    print("cell type not found")
  }
}

pngfunc("Plots/QC2/feature.png", width = 9, height = 5)
par(mfrow=c(2,4))
par(mar = c(1,1,2,1))
for(i in types){
  plot(sce$TSNEx, sce$TSNEy, col = rgb[[i]], pch = 19, cex=0.5, 
       xlab= "", ylab="", xaxt="n", yaxt="n", main= i)
  }

invisible(dev.off())
include_graphics("Plots/QC2/feature.png", dpi = 300)

Session info

sessionInfo()
## R version 4.2.0 (2022-04-22)
## Platform: x86_64-pc-linux-gnu (64-bit)
## Running under: Ubuntu 24.04.1 LTS
## 
## Matrix products: default
## BLAS/LAPACK: /cvmfs/soft.ccr.buffalo.edu/versions/2023.01/easybuild/software/avx512/Compiler/gcc/11.2.0/flexiblas/3.0.4/lib/libflexiblas.so.3.0
## 
## locale:
##  [1] LC_CTYPE=en_US.UTF-8 LC_NUMERIC=C         LC_TIME=C           
##  [4] LC_COLLATE=C         LC_MONETARY=C        LC_MESSAGES=C       
##  [7] LC_PAPER=C           LC_NAME=C            LC_ADDRESS=C        
## [10] LC_TELEPHONE=C       LC_MEASUREMENT=C     LC_IDENTIFICATION=C 
## 
## attached base packages:
## [1] grid      stats4    stats     graphics  grDevices utils     datasets 
## [8] methods   base     
## 
## other attached packages:
##  [1] SingleR_2.0.0               celldex_1.8.0              
##  [3] dplyr_1.1.4                 RSpectra_0.16-2            
##  [5] kableExtra_1.4.0            knitr_1.38                 
##  [7] Matrix_1.6-5                circlize_0.4.14            
##  [9] ComplexHeatmap_2.14.0       scran_1.26.2               
## [11] scater_1.26.1               ggplot2_3.5.1              
## [13] scuttle_1.8.4               SingleCellExperiment_1.20.1
## [15] SummarizedExperiment_1.28.0 Biobase_2.58.0             
## [17] GenomicRanges_1.50.2        GenomeInfoDb_1.34.9        
## [19] IRanges_2.32.0              S4Vectors_0.36.2           
## [21] BiocGenerics_0.44.0         MatrixGenerics_1.10.0      
## [23] matrixStats_0.62.0         
## 
## loaded via a namespace (and not attached):
##   [1] AnnotationHub_3.6.0           BiocFileCache_2.6.1          
##   [3] systemfonts_1.1.0             igraph_1.3.1                 
##   [5] BiocParallel_1.32.6           digest_0.6.29                
##   [7] foreach_1.5.2                 htmltools_0.5.2              
##   [9] viridis_0.6.2                 fansi_1.0.6                  
##  [11] magrittr_2.0.3                memoise_2.0.1                
##  [13] ScaledMatrix_1.6.0            cluster_2.1.3                
##  [15] doParallel_1.0.17             limma_3.54.2                 
##  [17] Biostrings_2.66.0             svglite_2.1.3                
##  [19] colorspace_2.1-1              blob_1.2.3                   
##  [21] rappdirs_0.3.3                ggrepel_0.9.1                
##  [23] xfun_0.30                     crayon_1.5.1                 
##  [25] RCurl_1.98-1.6                jsonlite_1.8.0               
##  [27] iterators_1.0.14              glue_1.8.0                   
##  [29] gtable_0.3.6                  zlibbioc_1.44.0              
##  [31] XVector_0.38.0                GetoptLong_1.0.5             
##  [33] DelayedArray_0.24.0           BiocSingular_1.14.0          
##  [35] shape_1.4.6                   scales_1.3.0                 
##  [37] pheatmap_1.0.12               DBI_1.1.2                    
##  [39] edgeR_3.40.2                  Rcpp_1.0.8.3                 
##  [41] viridisLite_0.4.2             xtable_1.8-4                 
##  [43] clue_0.3-60                   dqrng_0.4.1                  
##  [45] bit_4.0.4                     rsvd_1.0.5                   
##  [47] metapod_1.6.0                 httr_1.4.2                   
##  [49] RColorBrewer_1.1-3            ellipsis_0.3.2               
##  [51] farver_2.1.2                  pkgconfig_2.0.3              
##  [53] sass_0.4.1                    dbplyr_2.1.1                 
##  [55] locfit_1.5-9.5                utf8_1.2.4                   
##  [57] labeling_0.4.3                tidyselect_1.2.1             
##  [59] rlang_1.1.4                   later_1.3.0                  
##  [61] AnnotationDbi_1.60.2          munsell_0.5.1                
##  [63] BiocVersion_3.16.0            tools_4.2.0                  
##  [65] cachem_1.0.6                  cli_3.6.3                    
##  [67] generics_0.1.3                RSQLite_2.2.12               
##  [69] ExperimentHub_2.6.0           evaluate_0.15                
##  [71] stringr_1.5.1                 fastmap_1.1.0                
##  [73] yaml_2.3.5                    bit64_4.0.5                  
##  [75] purrr_1.0.2                   KEGGREST_1.38.0              
##  [77] sparseMatrixStats_1.10.0      mime_0.12                    
##  [79] xml2_1.3.3                    compiler_4.2.0               
##  [81] rstudioapi_0.13               beeswarm_0.4.0               
##  [83] filelock_1.0.3                curl_4.3.2                   
##  [85] png_0.1-7                     interactiveDisplayBase_1.36.0
##  [87] tibble_3.2.1                  statmod_1.4.36               
##  [89] bslib_0.3.1                   stringi_1.7.6                
##  [91] highr_0.9                     lattice_0.20-45              
##  [93] bluster_1.8.0                 vctrs_0.6.5                  
##  [95] pillar_1.9.0                  lifecycle_1.0.4              
##  [97] BiocManager_1.30.25           jquerylib_0.1.4              
##  [99] GlobalOptions_0.1.2           BiocNeighbors_1.16.0         
## [101] bitops_1.0-7                  irlba_2.3.5                  
## [103] httpuv_1.6.5                  R6_2.5.1                     
## [105] promises_1.2.0.1              gridExtra_2.3                
## [107] vipor_0.4.5                   codetools_0.2-18             
## [109] assertthat_0.2.1              rjson_0.2.21                 
## [111] withr_3.0.2                   GenomeInfoDbData_1.2.9       
## [113] parallel_4.2.0                beachmat_2.14.2              
## [115] rmarkdown_2.14                DelayedMatrixStats_1.20.0    
## [117] Rtsne_0.16                    shiny_1.7.1                  
## [119] ggbeeswarm_0.6.0